Increased COX6A2 Promotes Pancreatic β-Cell Apoptosis and Is Suppressed in Diabetic GK Rats After Roux-en-Y Gastric Bypass

Pancreatic β-cell apoptosis plays a key role in the development of type 2 diabetes (T2D). Bariatric surgery, such as Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy, has been demonstrated to effectively treat T2D and improve β-cell function (1–3). RYGB has been shown to suppress β-cell apoptosis (4,5), which contributes to the increased β-cell mass and subsequent remission of T2D (3). However, the mechanisms underlying improved β-cell apoptosis after RYGB remain unclear.

Glucagon-like peptide 1 (GLP-1) is an incretin released from enteroendocrine L cells. It amplifies the glucose-stimulated insulin secretion (GSIS) of β-cells via activation of intracellular cAMP/cAMP-dependent protein kinase (PKA) and cAMP/Epac2 signaling pathway (6). GLP-1 has also been reported to protect β-cells from apoptosis under diabetic conditions (7). We and others have demonstrated that RYGB increases the plasma GLP-1 level (8) and islet GLP-1R expression (9,10), suggesting that RYGB enhances the GLP-1 signaling pathway. However, whether the enhanced GLP-1 signaling is involved in RYGB-decreased β-cell apoptosis is largely unknown.

Carbohydrate response element binding protein (ChREBP, also known as Mlxipl), a glucose-sensing transcription factor, is expressed in multiple tissues, including pancreatic islets (11). ChREBP comprises two isoforms, ChREBPα and ChREBPβ (hereafter, ChREBP refers to ChREBPα). Upon exposure to high glucose, ChREBP is activated and mediates glucose-induced gene expression and β-cell apoptosis (12). After activation, ChREBP translocates to the nucleus and binds to the carbohydrate response element (ChoRE) on the target gene promoter, thereby recruiting coactivator, such as histone acetyltransferase p300, leading to enhanced histone acetylation at the target gene promoter and culminating in elevated target gene transcription (13). The transcriptional activity of ChREBP is contingent on its nuclear localization controlled by the phosphorylation of specific amino acid residues, such as Ser196, Ser568, and Thr666, which are phosphorylated by PKA and AMPK (14).

Cytochrome c oxidase subunit 6A2 (COX6A2) is expressed in pancreatic β-cells (15). Previous studies have shown that COX6A2 regulates cell apoptosis-related reactive oxygen species generation in skeletal muscles (16) and neurons (17). Deficiency of Cox6a2 protects mice against high-fat diet–induced obesity and insulin resistance (16). Currently, the role of COX6A2 in β-cell apoptosis, especially whether it is involved in RYGB-decreased β-cell apoptosis, is still elusive. Here we show that β-cell apoptosis and COX6A2 expression are increased in diabetic Goto-Kakizaki (GK) rat islets, whereas both are inhibited following RYGB surgery. Increased COX6A2 promotes β-cell apoptosis by modulating cyclophilin D (CypD)-mediated release of cytochrome c from mitochondria. Additionally, COX6A2 expression is epigenetically controlled by ChREBP. RYGB enhances GLP-1 signaling to phosphorylate and deactivate ChREBP, thereby inhibiting COX6A2 expression and subsequently alleviating β-cell apoptosis.

Male Wistar and GK rats (aged 8 weeks) were procured from SLRC (Shanghai, China), and 6-week-old male Sprague Dawley (SD) rats came from the Guangdong Medical Laboratory Animal Center. The GK and SD rats underwent Roux-en-Y gastric bypass (RYGB) or sham surgery. Additionally, some GK rats were given intraperitoneal injections of exendin-4 (10 μg/kg body wt/day) (18) or saline for 8 days. Cox6a2 knockout (KO) mice (C57BL/6JGpt) were generated using clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) technology to delete exons 1 to 3 in the Cox6a2 gene and were obtained from GemPharmatech Company (Nanjing, China). Because estrogen regulates the expression of COX6A2 (19), to avoid the interference of estrogen, male wild-type (WT) and Cox6a2 KO mice (aged 5 weeks) were used in this study. These mice were fed a high-fat diet for 3 months, followed by intraperitoneal injection of streptozotocin (STZ) (30 mg/kg body wt/day) (20) for 3 consecutive days to induce a diabetic state. The animal procedures followed the principles of laboratory animal care and were approved by the Shenzhen University Animal Care Committee. The pancreata of the rats and mice were processed by sectioning or by enzymatic digestion using collagenase P to isolate islets for examination.

The adeno-associated virus (AAV) AAV9-Cox6a2-KO plasmid was generated by inserting the single guide (sg)RNA (5′-TAAGGTCCTCAGTCGGAGCA-3′) target against the rat Cox6a2 gene into pX601-AAV-CMV-SaCas9-3xHA-U6-sgRNA vector. The AAV was packaged and provided by Vigene Biosciences (Jinan, China). To decrease COX6A2 expression in islets, GK rats were anesthetized with inhalation of 2–3% isoflurane and injected with 50 μL of AAV containing ∼2 × 10¹² Cox6a2-KO virus particles via the common bile duct. The control rats were injected with 50 μL of AAV control virus particles containing a nonspecific sgRNA. All rats were sacrificed 2 weeks later for biochemical analysis.

The RYGB and sham procedures were performed as previously described (21). Briefly, the rats were anesthetized with isoflurane after an overnight fast. The jejunum was detached at a distance of 16 cm from the Treitz ligament. The proximal segment was anastomosed to the 30 cm of the Roux limb, and the distal portion was stitched to the residual stomach. Some GK-RYGB rats received daily intraperitoneal injections of exendin(9-39) (30 μg/kg body wt) (22) or saline for 10 days.

The intravenous glucose tolerance test was performed as described previously (21). In brief, after an intravenous injection of 1.5 g/kg glucose, blood samples were taken from the jugular veins of rats to determine plasma insulin levels.

After overnight fasting, the mice were intraperitoneally injected with glucose (2 g/kg). Blood glucose levels were measured at 0, 15, 30, 60, 90, and 120 min after glucose administration.

INS-1 832/13 cells were cultured in RPMI 1640 medium (23). Rat Cox6a2 or CypD coding sequences were amplified by PCR using the primers listed in Supplementary Table 1, and the PCR products were inserted into the retroviral pMX-puro vector. INS-1 832/13 cells were transduced with pMX-puro-vector, pMX-puro-COX6A2, or pMX-puro-CypD, followed by selection with 3 μg/mL puromycin for 1 week. For the knockdown of ChREBP or COX6A2, INS-1 832/13 cells were transduced with either scramble or shChREBP (cat. RMM3981-201916728) or shCOX6A2 (cat. RMM3981-201787879) plasmid from GE Dharmacon.

The experiments were performed as previously described (23). The following antibodies were used: phospho-ChREBP (Ser196) (Proteintech), ChREBP (NB400-135, Novus), COX6A2 (11421-1-AP, Proteintech), cytochrome c (no. 11940, Cell Signaling Technology), COX IV (no. 4850, Cell Signaling Technology), cleaved caspase-3 (no. 9661, Cell Signaling Technology), caspase-9 (no. 9508, Cell Signaling Technology), CypD (ab110324, Abcam), GAPDH (no. 5174, Cell Signaling Technology), histone H3 (ab1791, Abcam), and β-actin (A5441, Sigma-Aldrich). The densities of the immunoblot bands were determined by Gel-Pro Analyzer 4.0 software.

RNA was extracted from INS-1 832/13 cells and islets using the Trizol method. Quantitative PCR analyses were conducted using the SYBR Green master mix from Promega on the ABI QuantStudio 5 system. Primer sequences used can be found in Supplementary Table 2. The quantity of target gene mRNA was normalized to GAPDH or β-actin.

Rat promoter fragments of Cox6a2 (−185 base pair [bp] to +65 bp) were amplified by PCR using primers shown in Supplementary Table 1 and subsequently cloned into a pGL3-basic plasmid (Promega). Site-directed mutagenesis of the putative ChREBP binding site from −20 bp to −15 bp in the Cox6a2 promoter was accomplished using the QuikChange II Mutagenesis Kit (Agilent Technology) with the primers listed in Supplementary Table 1. For determination of Cox6a2 promoter activity, human embryonic kidney 293 T cells were cotransfected with pGL3-Cox6a2 plasmid and pcDNA3.1 or ChREBP-overexpressing plasmid (no. 39235, Addgene), along with Renilla luciferase plasmid. After 48 h, the resulting promoter activity was measured using the Dual-Luciferase Reporter Assay kit (Promega).

Chromatin immunoprecipitation (ChIP) assays were performed as previously reported (21). Chromatin was immunoprecipitated with antibodies (3 μg) against ChREBP (NB400-135, Novus), acetylated histone H3 (06-599, Sigma-Aldrich) and p300 (ab14984, Abcam), respectively. The obtained DNA fragments were quantified by qPCR using the primers listed in Supplementary Table 3.

INS-1 832/13 cells were lysed in immunoprecipitation lysis buffer containing protease inhibitors. Cell lysates (1 mg) were incubated with 3 μg of CypD antibody overnight at 4°C. The complexes were immunoprecipitated with 60 μL protein A agarose beads for 3 h at 4°C and then washed sequentially with low-salt wash buffer, high-salt wash buffer, LiCl wash buffer, and Tris-HCl/EDTA buffer. The immunoprecipitated supernatants were separated by SDS-PAGE, followed by immunoblotting with COX6A2 antibody, and detected using the Tanon 5200 image system.

To determine the colocalization of COX6A2 and CypD, INS-1 832/13 cells were fixed with paraformaldehyde and permeabilized with Triton X-100. The cells were then incubated overnight at 4°C with primary antibodies against COX6A2 and CypD, followed by incubation with fluorescent secondary antibodies. Images were captured using a Zeiss LSM 510 confocal laser scanning microscope.

Additionally, pancreatic sections were stained with cleaved caspase-3, insulin, and DAPI. The resulting fluorescence was visualized using the Zeiss Axio Observer 3 fluorescent microscope. The insulin-positive and DAPI-stained areas were calculated using ImageJ software. β-Cell area was expressed as the ratio of the insulin-positive area to the DAPI-stained area. β-Cell mass was calculated through the β-cell area multiplied by the corresponding pancreatic weight.

The mitochondrial transmembrane potential (ΔΨm) of INS-1 832/13 cells was analyzed by JC-1 Mitochondrial Membrane Potential Assay Kit (Beyotime Biotechnology, Shanghai, China). The cells were incubated with JC-1 (1:200) dye for 20 min at 37°C. The cell-associated fluorescence was measured with an Axio Observer 3 microscope.

TUNEL assays were performed using a TUNEL assay kit (Beyotime Biotechnology). The samples were counterstained with DAPI to visualize the cell nucleus. Additionally, pancreatic sections were immunoassayed with anti-insulin antibodies to identify β-cells. The samples were observed using an Axio Observer 3 microscope. TUNEL-positive cells were counted and normalized to total β-cells.

INS-1 832/13 cells were collected after digestion with 0.25% trypsin. The cells were incubated with a mixture of propidium iodide and annexin V (Beyotime, Shanghai, China) for 20 min at room temperature, followed by immediate examination using flow cytometry.

Isolated islets were prepared for infection by adding 8 μg/mL polybrene. The islets were infected with adenovirus (Ad)-green fluorescent protein (GFP)-COX6A2 or Ad-GFP adenovirus (Sangon Biotech, Shanghai, China) at a multiplicity of infection of 500 for 8 h. Then the islets were transferred to a fresh medium. After 72 h of culture, the islets were harvested for Western blotting analysis.

The statistical analysis was performed using SPSS 20.0 software. The independent t test was used for comparing two groups, and one-way ANOVA with the least significant difference post hoc test was used for comparing more than two groups. A significance level of P < 0.05 was considered statistically significant.

The data sets generated during the current study are available upon request. The resources are available upon request.

The level of cleaved caspase-3 was examined 1 month after surgery to monitor islet cell apoptosis. As shown in Fig. 1A, RYGB markedly decreased apoptosis in islet cell populations in GK rats, consistent with previous reports (4,24). Correspondently, RYGB increased insulin-positive cell area (Fig. 1B and C), improved GSIS (Fig. 1D), and led to the remission of hyperglycemia in GK rats (Fig. 1E). We next determined the effect of RYGB on COX6A2 expression. Compared with Wistar islets, GK rat islets displayed higher COX6A2 mRNA and protein levels, respectively (Fig. 1F and G). Notably, the increased expression of COX6A2 was inhibited by RYGB (Fig. 1F and G). Comparable results were also obtained in SD-RYGB rats (Fig. 1H and I), indicating that RYGB could suppress COX6A2 expression in rat islets.

To investigate the role of increased COX6A2 in β-cell apoptosis, mouse islets were infected with Ad-GFP or Ad-GFP-COX6A2 to enhance COX6A2 expression. The infection with Ad-GFP-COX6A2 resulted in increased COX6A2 expression (Supplementary Fig. 1A) and an elevated level of cleaved caspase-3 (Fig. 2A). Similar results were also made in COX6A2-overexpressing INS-1 832/13 cells (Fig. 2Band Supplementary Fig. 1B). The increased apoptosis in COX6A2-overexpressing cells was also confirmed by TUNEL staining (Fig. 2C and D) and flow cytometry (Fig. 2E and F), respectively.

To further explore the role of COX6A2 in cell apoptosis, we also generated the scramble and COX6A2 knockdown (shCOX6A2) INS-1 832/13 cells (Supplementary Fig. 1C) and cultured these cells in 5.5 mmol/L glucose (G5.5) or 30 mmol/L glucose (G30) medium for 4 days. Consistent with the previous study (25), G30 culture enhanced cleaved caspase-3 levels in scramble cells (Fig. 2G). Of note, shCOX6A2 markedly reduced the increase of cleaved caspase-3 induced by high glucose (Fig. 2G). Thus, apoptosis caused by G30, as determined by the TUNEL assay, was blocked by shCOX6A2 (Fig. 2H and I). Similarly, COX6A2 deficiency diminished the STZ-induced cell apoptosis (Supplementary Fig. 2A and B).

To investigate the effect of COX6A2 on diabetic β-cell apoptosis in vivo, the WT and Cox6a2 KO mice were fed a high-fat diet and treated with STZ. The absence of COX6A2 expression in islets from Cox6a2 KO mice confirmed the knockout, as opposed to its detection in WT mice (Fig. 3A). Intriguingly, the level of cleaved caspase-3 was reduced in the KO islets compared with the WT islets (Fig. 3A and B). These data suggest that increased COX6A2 accelerates diabetic islet cell apoptosis. In line with the decreased apoptosis in Cox6a2 KO islets, the insulin-positive cell area was increased in KO mice (Fig. 3C and D). Consequently, Cox6a2 KO mice exhibited enhanced insulin levels (Fig. 3E and F), improved glycemic control (Fig. 3G and H), and reduced blood glucose levels (Fig. 3I).

To confirm the phenotype observed in the Cox6a2 KO mice, we further investigated the involvement of COX6A2 in islet cell apoptosis in vivo by delivering the AAV9-Cox6a2-KO (Cox6a2-AAV) virus or the control virus (vehicle) to diabetic GK rats. The administration of the AAV9-Cox6a2-KO virus resulted in decreased islet COX6A2 protein level (Supplementary Fig. 3A), which was accompanied by reduced cleaved caspase-3 and cleaved caspase-9 levels and TUNEL-positive and insulin-positive cells in GK rats (Fig. 4A–C and Supplementary Fig. 3B–D). Accordingly, the insulin-positive cell area (Fig. 4D) and mass (Fig. 4E) in GK-Cox6a2-AAV rats were higher than those of GK-vehicle rats, respectively. Therefore, GK-Cox6a2-AAV rats maintained a lower blood glucose level (Fig. 4F), suggesting an improvement in diabetes.

Given that COX6A2 is one of the subunits of mitochondrial complex IV (16), we examined the effect of COX6A2 overexpression on the ΔΨm. The vector control cells exhibited a higher ΔΨm, as indicated by stronger red and weaker green fluorescence. In contrast, COX6A2-overexpressing INS-1 832/13 cells displayed elevated green fluorescence, indicating a decrease in ΔΨm (Fig. 5A). Consequently, overexpression of COX6A2 promoted the release of cytochrome c from mitochondria to the cytoplasm (Fig. 5B) and hence activation of caspase-9 (Fig. 5C).

CypD, a constitutive subunit of the mitochondrial permeability transition pore (mPTP), has been shown to mediate the release of cytochrome c from mitochondria (26,27). We hypothesized that CypD could participate in COX6A2-induced cytochrome c release. We, therefore, investigated the interaction between CypD and COX6A2. The immunofluorescence results showed that CypD colocalized with COX6A2 in INS-1 832/13 cells (Fig. 5D). Coimmunoprecipitation data confirmed the association between CypD and COX6A2, which was enhanced by COX6A2 overexpression (Fig. 5E). Similar observations were also made in high-glucose cultured INS-1 832/13 cells (Fig. 5F). We further examined the effect of COX6A2 on CypD expression. Notably, overexpression of COX6A2 resulted in elevated CypD protein expression (Fig. 5G), albeit it did not affect CypD mRNA level (Supplementary Fig. 4); consistently, CypD protein level was enhanced in GK-vehicle rat islets (Fig. 5H). However, the increased CypD protein level was markedly reduced in GK-Cox6a2-AAV rat islets (Fig. 5H). Similar observations were also made in GK-RYGB rat islets (Fig. 5I). These findings reveal that COX6A2 positively regulates CypD protein expression. Subsequently, the crucial role of CypD in cell apoptosis was confirmed by increased caspase-3 activation in CypD-overexpressing cells (Fig. 5J). Furthermore, inhibition of CypD by cyclosporin A (CsA) suppressed cytochrome c release from mitochondria to the cytoplasm induced by COX6A2 overexpression (Fig. 5K and L), thereby reducing the increased level of cleaved caspase-3 in COX6A2-overexpressing INS-1 832/13 cells (Fig. 5M), suggesting that CypD mediates the effects of COX6A2 on β-cell apoptosis.

To ascertain the mechanism behind the increased COX6A2 expression in islets from diabetic rats, we examined the impact of high glucose on COX6A2 expression. In line with the elevated expression of COX6A2 in GK islets, G30 led to increased COX6A2 mRNA (Fig. 6A) and protein (Fig. 6B) expression, respectively. ChREBP, which is known to be activated by high glucose, mediates gene expression changes induced by such conditions (12). We then explored the role of ChREBP in regulating COX6A2 expression. The results showed that transiently ectopic expression of ChREBP enhanced both mRNA (Fig. 6C) and protein (Fig. 6D) expression of COX6A2. On the contrary, ChREBP knockdown suppressed COX6A2 expression (Fig. 6E). To gain further understanding of the mechanism by which ChREBP regulates COX6A2 expression, we analyzed the DNA sequence in the Cox6a2 locus and identified a “CAGGTG” sequence as the putative ChREBP binding site (ChoRE) within the Cox6a2 promoter. We then constructed a luciferase reporter driven by the WT ChoRE or a mutant form “AGTTCT” (Fig. 6F) and assessed the luciferase activity. The data demonstrated that ChREBP increased reporter expression driven by the WT ChoRE but not by the mutant form (Fig. 6F), signifying that the ChoRE in the Cox6a2 promoter is essential for ChREBP-regulated COX6A2 expression. Moreover, the ChIP assay revealed that high glucose promoted ChREBP binding to the promoter of Cox6a2 (Fig. 6G) and increased the acetylation of histone H3 (ACH3), a marker associated with gene activation, at this promoter region (Fig. 6H). Considering that histone acetyltransferase p300 is a key enzyme facilitating ACH3 formation (28), we posited that p300 is involved in ACH3 production at the Cox6a2 promoter. Supporting this hypothesis, ChIP assays demonstrated an increased occupancy of p300 at the Cox6a2 promoter under high-glucose conditions (Fig. 6I). Notably, inhibition of p300 with its specific inhibitor C646 mitigated the aberrant expression of COX6A2 induced by high glucose (Fig. 6J), confirming that p300 mediates this process. These findings suggest that high glucose activates ChREBP, which then recruits p300 to enhance ACH3 at the Cox6a2 promoter, thereby upregulating COX6A2 expression in β-cells.

Given the critical role of ChREBP in mediating COX6A2 expression, we evaluated the impact of RYGB on ChREBP phosphorylation. As shown in Fig. 6K, GK-sham rats displayed decreased ChREBP phosphorylation. However, RYGB led to enhanced ChREBP phosphorylation, suggesting an inhibitory effect of RYGB on ChREBP. As previous research has shown increased GLP-1 levels in patients after RYGB (8), and since PKA can phosphorylate ChREBP (14), we hypothesize that RYGB recovered ChREBP phosphorylation via an enhanced GLP-1 signaling pathway. To this end, we confirmed the increased GLP-1 levels in GK-RYGB rats (Supplementary Fig. 5A). Subsequently, we determined the stimulatory effect of GLP-1 signaling on ChREBP phosphorylation. As expected, the GLP-1R agonist exendin-4 or adenylate cyclase activator forskolin markedly promoted the phosphorylation of ChREBP (Fig. 7A and B), which was abolished by the specific PKA inhibitor Rp-8-Br-2’-O-MB-cAMPs (Rp-cAMP) (Fig. 7B). Given that the phosphorylation of ChREBP controls its nuclear translocation (14), we investigated the effect of exendin-4 on ChREBP subcellular distribution. Consistent with a previous study (29), high glucose promoted ChREBP translocation from the cytoplasm to the nucleus, which was inhibited by exendin-4 (Fig. 7C and D). Accordingly, exendin-4 abrogated high glucose–increased ACH3 at the Cox6a2 promoter (Fig. 7E), thereby diminishing high glucose–induced COX6A2 expression in a PKA-dependent manner (Fig. 7F and G and Supplementary Fig. 5B and C). Similar observations were also made in GK rats intraperitoneally injected with exendin-4 (Fig. 7H). As expected, exendin-4 treatment led to an increased insulin-positive cell area and alleviated hyperglycemia in GK rats (Supplementary Fig. 6A–C).

To further elucidate the role of the GLP-1 signaling pathway in the RYGB-decreased COX6A2 expression, GK-RYGB rats were intraperitoneally injected with the GLP-1R antagonist exendin(9-39) or saline. Notably, exendin(9-39) treatment resulted in increased expression of COX6A2 (Fig. 7I) and an elevated level of cleaved caspase-3 (Fig. 7J), alongside reduced insulin-positive cell area (Fig. 7K) within the islets of GK-RYGB rats compared with the saline-treated group. Consequently, rats treated with exendin(9-39) displayed lower plasma insulin levels (Fig. 7L) and higher blood glucose levels (Fig. 7M) compared with the saline control rats. These findings underscore the pivotal role of GLP-1 signaling in the RYGB-mediated inhibition of COX6A2 expression and β-cell apoptosis.

In this study, we demonstrate that RYGB reduces aberrant COX6A2 expression in islets from GK rats. The expression of COX6A2 is modulated by the GLP-1/PKA/ChREBP axis via a p300-mediated increase in ACH3 at the Cox6a2 promoter. For the first time, we show the proapoptotic effect of increased COX6A2 in β-cells through modulation of CypD (Fig. 8).

The mitochondria-dependent cell apoptosis is characterized by decreased mitochondrial membrane potential, increased permeability of the mitochondrial membrane via opening mPTP, and release of cytochrome c under the stimulation of apoptotic signals. This then triggers apoptotic cascades, ultimately leading to apoptosis (30). Here, we demonstrated the crucial role of COX6A2 in promoting diabetic β-cell apoptosis, supported by the decreased cleaved caspase-3 and TUNEL immunofluorescence under suppressed or depleted COX6A2 levels. These observations may suggest that decreased β-cell apoptosis mediated by suppressed COX6A2 expression contributes to improved β-cell function, as indicated by increased β-cell mass, thus alleviating diabetic symptoms of hyperglycemia. Surprisingly, COX6A2 can interact with CypD. Moreover, CypD expression was positively correlated with COX6A2 expression. In the conditions where COX6A2 was overexpressed, there was higher CypD, higher cytochrome c release into the cytosol, and increased cleaved caspase-3 level. CypD plays a key role in mPTP function and promotes cell apoptosis via cytochrome c release, while inhibition suppresses cell apoptosis (26,31,32). The failure of COX6A2 overexpression to induce cytochrome c release and caspase-3 activation in the presence of CypD inhibitor CsA emphasized the essential role of CypD in COX6A2-mediated β-cell apoptosis. However, the molecular mechanism underlying COX6A2-induced CypD protein expression requires further exploration. Furthermore, given the constraints of in vivo models, further investigation is necessary to ascertain whether COX6A2 directly modulates β-cell apoptosis in vivo.

ChREBP is a glucose-response transcription factor that mediates glucose-induced gene expression. Exposure of β-cells to high glucose leads to an increase in ChREBP levels (33), and strikingly, increased ChREBP triggers β-cell apoptosis (12). However, the downstream target genes involved in ChREBP-mediated cell apoptosis remain largely unidentified. For the first time, we demonstrated that COX6A2 is one of the target genes regulated by ChREBP in β-cells. Three pieces of evidence support this view. First, activation or overexpression of ChREBP led to increased COX6A2 expression. Second, ChREBP knockdown suppressed COX6A2 expression. Third, there exists the ChoRE at the Cox6a2 promoter, and stimulation with high glucose resulted in increased ChREBP binding to the Cox6a2 promoter. Furthermore, our findings also revealed that activation of ChREBP by high glucose promoted the recruitment of the epigenetic regulator p300 and subsequently enhanced histone H3 acetylation at the Cox6a2 locus. This would lead to increased expression of COX6A2, given that acetylation of histones promotes gene transcription (28). Consistently, the p300 inhibitor C646 alleviated high glucose–induced COX6A2 expression, indicating that p300 is at least partially responsible for these effects of ChREBP. Therefore, the ChREBP/p300/COX6A2 axis may play a fundamental role in diabetic β-cell apoptosis.

Our data also demonstrated that RYGB downregulated COX6A2 expression in the islets of both diabetic GK rats and healthy SD rats. This downregulation would be a result of RYGB-induced inactivation of ChREBP through promoting ChREBP phosphorylation. RYGB has been proven to augment GLP-1 signaling (8–10). Correspondently, exendin-4 or forskolin phosphorylated ChREBP in a PKA-dependent manner. Moreover, the GLP-1R antagonist exendin(9-39) reversed the inhibitory effect of RYGB on COX6A2 expression. These data suggested that RYGB inhibits COX6A2 expression via modulation of the GLP-1/PKA/ChREBP signaling pathway. Therefore, RYGB likely reduces β-cell apoptosis by inhibiting COX6A2 expression.

In conclusion, our findings suggest that increased COX6A2 may facilitate β-cell apoptosis through the modulation of CypD-mediated cytochrome c release. The increase in COX6A2 expression could be a result of hyperglycemia-induced ChREBP overactivation in diabetes. RYGB is likely to reduce β-cell apoptosis by regulating the GLP-1/PKA/ChREBP/COX6A2 pathway. These findings not only uncover a novel way for regulating β-cell apoptosis but also highlight COX6A2 as a potential target for the treatment of T2D.

Funding. This work was funded by the National Natural Science Foundation of China (82070806, 82070845, 82370833, 82070978, 82072661), Natural Science Foundation of Guangdong Province, China (2023A1515010442), Shenzhen Science and Technology programs (JCYJ20210324094812033), and Shenzhen Key Laboratory of Metabolism and Cardiovascular Homeostasis (ZDSYS20190902092903237).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. X.K. analyzed data and wrote the manuscript. X.K., D.Y., L.S., B.L., S.L., Y.T., and Y.Z. performed the experiments. X.K., X.S., Y.Y., and X.M. designed the experiments. X.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.